Information Aggregation with AI Agents
econ.GN, cs.AI, cs.GT, q-fin.EC
Submitted: 2026-04-21
Updated: 2026-09-25
Comments: 80 pages
License: http://creativecommons.org/licenses/by/4.0/
The gist: Can Large Language Models (AI agents) aggregate dispersed private information through trading and reason about the knowledge of others by observing price movements? We conduct a controlled experiment
Terminology
Abstract
Can Large Language Models (AI agents) aggregate dispersed private information through trading and reason about the knowledge of others by observing price movements? We conduct a controlled experiment where AI agents trade in a prediction market after receiving private signals, across four information structures of increasing complexity. We find that although the median market is effective at aggregating information in the easy information structures, performance deteriorates in the harder structures, suggesting that AI agents struggle in environments where more than two levels of interactive reasoning are required, a ceiling close to the one documented in human subjects. Consistent with our theoretical predictions, market accuracy does not improve from allowing cheap talk communication, changing the duration of the market, or strategic prompting; initial price has little average effect but matters in the very hard structure. We also find that ``smarter'' AI agents perform better at aggregation and are more profitable. Surprisingly, giving them feedback about past performance does not improve aggregation. A further wave of markets, run three months later with capability-frontier models, aggregates information more often in the three easier structures but not in the hardest one, where higher capability replaces markets that are confidently wrong with markets that hedge near 0.5.
Sources
- $\tau^2$-Bench: Evaluating Conversational Agents in a Dual-Control Environment
- Designing Informative Securities
- LiveCodeBench: Holistic and Contamination Free Evaluation of Large Language Models for Code
- Language Models (Mostly) Know What They Know
- Towards a Science of Scaling Agent Systems
- Humanity's Last Exam
- Generalizing Verifiable Instruction Following
- How manipulable are prediction markets?
- Can Large Language Models Express Uncertainty Like Human?
- Can LLMs Express Their Uncertainty? An Empirical Evaluation of Confidence Elicitation in LLMs
Related papers
- How optimistic inflow forecasts distort dispatch, prices, and contracts in hydro-dominated power systems: evidence from Brazil
- The Economics of AI Inference: Inflation Dynamics, Welfare Costs, and Optimal Monetary Policy under the Inference-Cost Phillips Curve
- AI Economist Agent: An Agentic Framework for Evidence-Based Economic and Financial Analysis with RAG, Knowledge Graphs, and Large Language Models
- The Economics of Model Collapse: Equilibrium, Welfare, and Optimal Provenance Subsidies in Synthetic Data Markets
- Is Decentralized Finance Actually Decentralized? An Interdisciplinary Framework Integrating Network Theory, Agent-Based Simulation, and Longitudinal Evidence from Aave, GHO Issuance, and Cross-Chain Expansion
- Delphos: A reinforcement learning framework for assisting discrete choice model specification